Support Systems Engineer

As our Support Systems Engineer, you'll own the systems behind Guardz's AI-first support: Guardy, our support agent; the support platform it runs on; and the AI agents that classify escalations, investigate issues, and draft fixes that make Guardy smarter.
You'll also be Support's Tier 4 escalation point, taking strategic customer bugs from root cause to a tested, merged fix.

As the first hire in this role, you'll report to the Director of Support.

Responsibilities:

• Support AI Performance: Own Fin Guidance, Procedures, Data Connectors and escalation rules, batch-testing and simulating every change before it goes live, and own the Support AI roadmap.
• Knowledge Architecture: Own the Help Center and internal knowledge as one structured source of truth for people and AI, kept current with every release.
• Knowledge Review Gate: Approve every knowledge change our agents propose until each class of change earns the right to ship on its own.
• Intercom Administration: Own Intercom workspace configuration, permissions, ticket types, attributes and workflows.
• Integrations: Build and maintain integrations through the Intercom API and webhooks, connecting Fin to the Guardz platform, Linear, Slack, HubSpot and our data warehouse.
• Agent Operations: Operate and extend the Support AI agents infrastructure built around Intercom, keeping them scheduled, monitored and cost-aware, with prompts and skills managed as versioned code
• Evaluation & Trust: Build the evaluation sets that prove each agent works, set trust thresholds per intent class, and own the safety rails: audit log, rollback and kill switches.
• New Capabilities: Partner with Engineering to let Guardy gather customer context and platform configurations, and empower it to use this context to resolve a greater variety of requests.
• Strategic Bug Resolution (T4): Be Support's final escalation point for high-impact customer bugs, owning them from reproduction and root cause through code, tests, pull request and production verification.
• AI-Assisted Fixes: Validate agent-generated RCAs and pull requests and drive them to merge with the owning team, who review every change. Hand back fixes that outgrow a few days' scope with findings and a proposed approach.
• Supportability: Ship changes that prevent repeat tickets: clearer errors, better logging and safer retries.

Requirements:

• 4+ years in support, escalation or software engineering, CX operations or automation engineering, with 2+ in B2B SaaS.
• Hands-on helpdesk administration (Intercom, Plain or Pylon preferred; Zendesk counts): workflows, data model, APIs.
• LLM-powered workflows or agents you've run in production, including prompt and skill design, tool use and evaluations.
• TypeScript/Node.js or Python, SQL (PostgreSQL), REST APIs, webhooks, Git, pull requests and automated tests.
• Proven ability to debug production systems you didn't write, using logs, traces and database queries
• Clear structured writing: a help article, a Fin Guidance rule and a technical spec, equally well.
• A habit of measurement: you baseline a metric, change something, and show whether it moved.
• Working knowledge of IT security and MSP environments (Microsoft 365, Google Workspace, identity, endpoint, email).
• Excellent communication skills in English, and the ability to work independently as the owner of a function.